基于无线传感器网络的目标定位和跟踪算法研究

发布时间：2018-02-28 03:48

本文关键词： 无线传感器网络 RSSI测距目标定位非线性滤波机动目标跟踪　出处：《太原理工大学》2017年硕士论文　论文类型：学位论文

【摘要】：无线传感器网络(Wireless Sensor Networks,WSN)作为一种新兴的交叉学科,集成了微电子、无线通讯、数据传输等多种技术,在各领域均有着较为普遍的运用前景。近几年来,由于WSN具备体积小、成本低、能耗少等特殊性,在公共安全、智能生活、军事防御等范畴有着突出的地位,成为当前研究的热点。目标定位和跟踪是WSN的一个较为重要的应用,由于目标的运动方式以非线性为主,因此要实现较好的跟踪有很大的困难。目前WSN目标跟踪算法存在跟踪精度不高,算法复杂度较大,实时性差等缺陷,如何实现准确、快速、稳定的目标跟踪成为广大研究者努力的方向。首先,以目标定位算法为切入点,研究了基于RSSI测距的三种目标定位算法,并指出算法性能的评价指标:精度和复杂度。由静止目标定位和运动目标跟踪的仿真可知:RSSI-TL算法有较好的精度和实时性,且跟踪过程比较稳定;RSSI-WC算法的精度仅次于RSSI-TL算法;RSSI-MLE算法在目标定位和跟踪过程中均产生了较大的误差,精度较差。其次,讨论了四种非线性滤波目标跟踪算法,分别对线性匀速运动目标和非线性变速运动目标进行仿真跟踪,对比分析MLKF、EKF、UKF和PF算法的跟踪性能。结果表明:在线性匀速运动目标跟踪过程中,PF的跟踪精度最高,EKF最差,其中MLKF的跟踪用时最短,而PF跟踪用时最长。同时随着节点个数的增加,MLKF、PF与UKF的跟踪精度逐渐提高,而EKF的跟踪精度呈现出先降低后升高的态势;在非线性变速运动目标跟踪过程中,MLKF的跟踪精度和实时性均优于其他算法,而PF的跟踪精度和实时性最差,并且传感器节点的个数对MLKF和PF算法跟踪精度的影响不明显。然后,分析了CV、CA、CT三种数学目标运动模型,引出了对IMMPF机动目标跟踪算法的研究,并对CV-CT目标运动模型进行仿真跟踪,对比分析了MLKF、PF与IMMPF算法的跟踪性能。仿真可知:IMMPF算法的跟踪精度明显高于PF、MLKF算法,有较好的实时性,且随着采样粒子个数的增加,PF与IMMPF算法的跟踪精度均有明显的提高,但跟踪用时亦有所增加。接着,针对IMMPF算法在跟踪过程中存在粒子退化现象进行了改进,在线性模块采用MLKF算法滤波,非线性模块采用改进残差重抽样的PF算法滤波,并使用加权平均输出目标的预测值。通过对CA-CV和CA-CT-CV模型的机动目标仿真跟踪,对比分析了MLKF、RSSI-TL、IMMPF和改进算法的跟踪性能,验证了改进算法的有效性。结果表明:改进算法的跟踪精度明显最高,而MLKF最差,其中RSSI-TL的跟踪用时最短,同时随着传感器节点个数的增加,改进算法的跟踪精度基本保持不变。最后,总结全文,指出在环境比较恶劣或目标运动较复杂的情况下,实现准确、快速、稳定的目标跟踪还需进一步研究。
[Abstract]:Wireless Sensor Networks (WSNs), as a new interdisciplinary subject, integrates many technologies, such as microelectronics, wireless communication, data transmission and so on, and has a wide application prospect in various fields. In recent years, because of the small size of WSN, WSNs are widely used in many fields. Because of its low cost and low energy consumption, it has a prominent position in the fields of public security, intelligent life, military defense and so on, and has become a hot spot in current research. Target location and tracking is a relatively important application of WSN. It is very difficult to achieve better tracking because the motion mode of the target is mainly nonlinear. At present, the WSN target tracking algorithm has some defects such as low tracking precision, large complexity, poor real-time performance, etc., how to achieve the accuracy and speed, and so on. Stable target tracking has become the direction of researchers' efforts. Firstly, three target location algorithms based on RSSI ranging are studied by using target location algorithm as the starting point. The accuracy and complexity of the algorithm are pointed out. The simulation of stationary target location and moving target tracking shows that the algorithm has better accuracy and real-time performance. The tracking process is relatively stable and the precision of RSSI-WC algorithm is second only to that of RSSI-TL algorithm. The accuracy of RSSI-MLE algorithm is lower than that of RSSI-MLE algorithm. Secondly, four nonlinear filtering target tracking algorithms are discussed. The tracking performance of linear uniform moving target and nonlinear variable speed moving target is compared and analyzed. The results show that the tracking accuracy of PF is the highest than that of EKF in the process of linear uniform moving target tracking. The tracking time of MLKF is the shortest, and that of PF is the longest. With the increase of node number, the tracking accuracy of MLKF / PF and UKF increases gradually, while the tracking accuracy of EKF decreases first and then increases. In the process of nonlinear moving target tracking with variable speed, the tracking accuracy and real-time performance of MLKF are better than those of other algorithms, but the tracking accuracy and real-time performance of PF are the worst, and the number of sensor nodes has no obvious effect on the tracking accuracy of MLKF and PF algorithms. In this paper, three kinds of mathematical target motion models are analyzed, and the research of IMMPF maneuvering target tracking algorithm is introduced, and the simulation tracking of CV-CT target motion model is carried out. The simulation results show that the tracking accuracy of the IMMPF algorithm is obviously higher than that of the MLKF algorithm, and the tracking accuracy of the PF algorithm and the IMMPF algorithm are improved with the increase of the number of particles sampled. However, the tracking time is also increased. Then, the particle degradation phenomenon in the IMMPF algorithm is improved, the linear module is filtered by the MLKF algorithm, and the nonlinear module is filtered by the improved residual resampling PF algorithm. Using the prediction value of weighted average outputted target, the tracking performance of MLKFF RSSI-TLMMPF and the improved algorithm is analyzed by simulation tracking of maneuvering target in CA-CV and CA-CT-CV models. The effectiveness of the improved algorithm is verified. The results show that the tracking accuracy of the improved algorithm is the highest, and that of MLKF is the worst. The tracking time of RSSI-TL is the shortest, and the number of sensor nodes increases. The tracking accuracy of the improved algorithm remains basically unchanged. Finally, the paper summarizes the full text and points out that further research is needed to achieve accurate, fast and stable target tracking when the environment is relatively bad or the target motion is more complicated.
【学位授予单位】：太原理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP212.9;TN929.5

【相似文献】